Kepler and Axiom Normalize Orbital Procurement
Starcloud
The real risk is that Kepler and Axiom make buying space compute feel routine before Starcloud proves that giant orbital GPU clusters work reliably and cheaply. Kepler already has live hardware in orbit and a network to move data between satellites, while Axiom is packaging nodes as infrastructure that governments and enterprises can actually contract for. That shifts the early market from bold technical vision to who can close the first repeatable procurement motion.
-
Kepler is ahead on operational proof. It commissioned 40 NVIDIA Jetson Orin modules across 10 optical relay satellites in March 2026, then opened that cluster to customers in April. That means buyers can test workloads on an existing mesh instead of waiting for a new dedicated compute constellation.
-
Axiom is turning orbital compute into a familiar buying model. Its first two orbital data center nodes launched on January 11, 2026 inside Kepler’s network, and Axiom is positioning them for national security, commercial, and international customers. In practice, that looks more like leasing infrastructure from a landlord than betting on a frontier hardware buildout.
-
This matters because early space infrastructure markets often harden around compliance, interfaces, and incumbent relationships before peak technical performance arrives. Defense procurement experience shows that once a vendor is inside the workflow, follow on awards are easier. Starcloud has raised more capital for a larger pure play compute vision, but today it still has no revenue and only early technical proof.
The next phase is likely to split the market in two. Kepler and Axiom can win near term demand for in orbit processing, sovereign data handling, and defense pilots, while Starcloud aims at the much bigger prize of true cloud scale compute in space. If procurement habits form around networked nodes first, Starcloud will need to beat not just physics, but an already established buying pattern.